利用目标区域拓扑关系图提取建筑物点云

发布时间：2018-03-24 16:26

本文选题：机载激光雷达　切入点：点云数据　出处：《武汉大学学报(信息科学版)》2017年04期

【摘要】：建筑物提取一直是机载激光点云数据处理研究的热点,其中建筑物和其他地物之间的区分是研究的核心和难点。为提高建筑物与其他地物在机载激光点云中的区分能力,提出了一种建筑物点云层次提取方法。首先,在点云滤波后,从非地面点云中提取建筑物候选区域;然后,通过形态学重建和点云平面分割方法对建筑物候选区域构建多尺度空间,并建立目标区域的拓扑关系图;最后,在拓扑关系图基础上,利用5种特征量对目标区域分类,并精确提取建筑物点云。为了测试算法的有效性和可靠性,利用国际摄影测量与遥感学会(International Society for Photogrammetry and Remote Sensing,ISPRS)提供的Vaihingen和Toronto两组测试数据集进行实验,并由ISPRS对结果进行评估,其中基于面积和目标的完整度、正确率和提取质量分别都大于87.8%、94.7%、87.3%。与其他建筑物提取方法相比,该方法在基于面积和目标的质量指标方面最为稳定。实验结果表明,在不同的城市场景下,该算法能够稳健地提取建筑物,并保持很高的正确率。
[Abstract]:Building extraction has always been a hot topic in airborne laser point cloud data processing, in which the distinction between buildings and other features is the core and difficulty. In this paper, a method of building point cloud level extraction is proposed. Firstly, after point cloud filtering, building candidate area is extracted from non-ground point cloud, and then, By means of morphological reconstruction and point cloud plane segmentation, multi-scale space is constructed for candidate areas of buildings, and topological relation diagrams of target regions are established. Finally, based on topological relation diagrams, five kinds of feature quantities are used to classify target regions. In order to test the validity and reliability of the algorithm, two sets of test data sets, Vaihingen and Toronto, provided by the International Society of Photogrammetry and remote Sensing (ISPRS), were used to test the results and the results were evaluated by ISPRS. Among them, based on the integrity of area and target, the accuracy and quality of extraction are all greater than 87.8% and 94.7%, respectively. Compared with other methods of building extraction, this method is the most stable in terms of quality index based on area and target. The experimental results show that, In different urban scenarios, the algorithm can extract buildings stably and maintain a high accuracy.
【作者单位】：武汉大学测绘遥感信息工程国家重点实验室;中国地质大学(武汉)信息工程学院;
【基金】：国家科技支撑计划(2014BAL05B07) 国家自然科学基金(41531177) 海洋公益性行业科研专项经费(2013418025-6)~~
【分类号】：P23

【相似文献】